Characterization of patient-derived tumor xenograft models of endometrial cancer for preclinical evaluation of targeted therapies.

ABSTRACT

OBJECTIVE: Endometrial carcinoma (EC) is the sixth most common cancer in women and therapies are limited for advanced and recurrent disease. Patient-derived tumor xenograft (PDTX) models are becoming popular tools in translational research because of their histological and genetic similarity to the original tumors and the ability to predict therapeutic response to treatments. Here, we established and characterized a panel of 24 EC PDTX models which includes the major histological and genetic subtypes observed in patients. METHODS: Fresh tumor tissues collected from primary, metastatic and recurrent type I and type II EC patients were engrafted in immunocompromised mice. Histology, vimentin, and cytokeratin expression were evaluated, together with Microsatellite instability (MSI), mutation profiling by Whole Exome Sequencing and copy number profiling by Whole Genome Low Coverage Sequencing. The efficacy of both PI3K and MEK inhibitors was evaluated in a model of endometrioid carcinoma harboring PTEN, PIK3CA and KRAS mutations. RESULTS: We observed good similarity between primary tumors and the corresponding xenografts, at histological and genetic level. Among the engrafted endometrioid models, we found a significant enrichment of MSI and POLE mutated tumors, compared to non-engrafted samples. Combination treatment with NVP-BEZ235 and AZD6244 showed the possibility to stabilize the tumor growth in one model originated from a patient who already received several lines of chemotherapy. CONCLUSION: The established EC PDTX models, resembling the original human tumors, promise to be useful for preclinical evaluation of novel combination and targeted therapies in specific EC subgroups.